I store HTML and various bits of text in a JSON structure in Postgresql, written using Python and the pg8000 library. I have found that I often get errors similar to the following:
pg8000.core.ProgrammingError: {'S': 'ERROR', 'V': 'ERROR', 'C': '22P02', 'M': 'invalid input syntax for type json', 'D': 'Escape sequence "\\&" is invalid.', 'W': 'JSON data, line 1: ...Ga,"'"));-1!=a.indexOf("\\&...', 'F': 'json.c', 'L': '951', 'R': 'json_lex_string'}
So to strip these I have the following horrifically slow code that I keep having to add specific 'invalid escape sequences' to:
def filter_values(value: Any) -> Any:
if isinstance(value, list):
return list(map(filter_values, value))
if isinstance(value, dict):
return {k:filter_values(v) for k, v in value}
if isinstance(value, str):
stripped = value.strip().replace('\\u0000', '').replace('\u0000', '').replace(u'\u0000', '')
while '\\-' in stripped:
stripped = stripped.replace('\\-', '-')
while '\\+' in stripped:
stripped = stripped.replace('\\+', '+')
while '\\E' in stripped:
stripped = stripped.replace('\\E', 'E')
while '\\&' in stripped:
stripped = stripped.replace('\\&', '&')
return stripped
return value
If I can't just strip all \ as that will break the json (such as \"). There has to be a better way to do this. What are the rules around correctly escaping characters in JSON Postgresql blobs?